Facts, Targeting, and the Difference Between Noise and Votes
What the research says — and what June 27 just demonstrated in real life. Four lessons from a clean election night in St. Tammany Parish.
Facts, Targeting, and the Difference Between Noise and Votes
What the research says — and what June 27 just demonstrated in real life
By Sam Caruso
Election night delivered clean results:
- St. Tammany Parish Mosquito Abatement District renewal: 69% yes
- Sheriff's Office renewal: 65% yes
- Slidell Prop 1: 69% yes
- Slidell Prop 2: 69% yes
The next morning I wrote one sentence in my notebook:
Facts still matter. Targeting matters. Internet noise is often just that — noise. And voter support tends to match the strength of the case you make.
1) Facts still matter (because voters respond to credible information)
When a campaign gets attacked, the temptation is to fight the vibe: rebut the loudest accusation, dunk on the worst critic, "win the internet."
That's usually the wrong game.
In New Haven's 1998 election, Gerber & Green ran a randomized field experiment and found that face-to-face canvassing increased turnout (their estimate was about six percentage points).
That doesn't mean "facts win" in some moral sense. It means campaigns that make the verifiable case and put it in front of the right people tend to do better than campaigns that treat persuasion like a shouting contest.
Takeaway: When you're under attack, keep returning to the documented case: what you do, what it costs, what voters get, and where the proof is.
2) Targeting matters (and the research explains why)
A June tax election is low turnout and low salience. Roughly 42,000 people decided parishwide millages — a small, self-selecting slice of registered voters.
Arceneaux & Nickerson reanalyzed 11 field experiments and argued (and found evidence consistent with the idea) that GOTV efforts tend to mobilize people who are near the "indifference threshold" — the voters who are on the edge of showing up. The key point for June: when election interest is low, the pool of "almost voters" is smaller and harder to reach.
So if you target badly in a low-salience election, you don't just waste money — you burn the limited number of contacts you have on people who were never going to vote anyway.
Takeaway: In sleepy elections, field programs are not "spray and pray." They're triage.
3) Noise isn't votes (and the platforms are built to confuse you)
If you had judged the Mosquito Abatement renewal by the comment sections, you would have predicted a loss. Opposition was vocal, organized, and constant. Support looked like one person saying, "I like being able to sit on my porch in July," and then going back to work.
That mismatch isn't mysterious once you accept one simple fact: engagement isn't public opinion. Platforms are optimized for what gets clicks, replies, and shares — not for what's representative or correct.
Rathje, Van Bavel, and van der Linden studied engagement on Facebook and Twitter at scale and found that out-group animosity was associated with higher engagement (virality). The incentive is obvious: posts that attack an out-group travel farther than posts that explain a budget line item.
So the internet becomes a funhouse mirror: the angriest content is what you see the most, which makes it feel like the angriest people are "the public." They're not.
Takeaway: Treat social media as a sensor for what's inflamed, not a measure of what will pass.
"One practical way to keep yourself honest during a campaign is to hold ongoing community meetings" (not just one "big" town hall), Kevin Caillouet, St. Tammany Parish Mosquito Abatement District. Kevin gets it. Community meetings, usually a lot of them, with smaller audiences, do two things that social platforms can't:
- Surface real objections early — the ones that show up when people have to raise a hand and ask.
- Show you who's actually in the room — which is often a better proxy for persuadable voters than the loudest online accounts.
My working assumption: the "social media warriors" mostly aren't in the room. Voters are.
Takeaway: If you want a read on what will pass, watch who shows up, what they ask, and what changes their mind.
4) Voter support is more consistent than people think
Look at the clustering:
| Measure | Result |
|---|---|
| Mosquito Abatement renewal | 69% |
| Slidell Prop 1 | 69% |
| Slidell Prop 2 | 69% |
| Sheriff's renewal | 65% |
Three separate campaigns, different agencies, same neighborhood of support.
That looks like a stable "yes" coalition doing what stable coalitions do: approving a renewal when the service is understood and the case is coherent.
The Sheriff's renewal running four points lower — weeks after a public scandal — also gives you a practical way to think about headwinds: sometimes controversy costs points; it does not automatically cost the election.
Takeaway: Your ceiling is not random. It's connected to how easy it is for a reasonable voter to say yes.
The through-line
The internet rewards anger.
Elections reward arguments you can explain, document, and repeat to the right voters.
Sources
[1] Alan S. Gerber & Donald P. Green, "Does canvassing increase voter turnout? A field experiment," Proceedings of the National Academy of Sciences (1999). PubMed abstract
[2] Kevin Arceneaux & David W. Nickerson, "Who Is Mobilized to Vote? A Re-Analysis of 11 Field Experiments," American Journal of Political Science 53(1): 1–16 (2009). PDF
[3] Steve Rathje, Jay J. Van Bavel, & Sander van der Linden, "Out-group animosity drives engagement on social media," PNAS 118(26): e2024292118 (2021). https://www.pnas.org/doi/10.1073/pnas.2024292118
Explore Topics
Written by
Sam Caruso
Content creator and writer sharing insights and stories.
